The concept of levels of organization in the biological sciences

(1)

The Concept of Levels of Organization in the Biological

Sciences

PhD Thesis

Submitted August 2014

Revised June 2015

Daniel Stephen Brooks

Department of Philosophy

Bielefeld University

Reviewers:

Martin Carrier, Maria

Kronfeldner

(2)

Dedicated in friendship to Jan and Magga

(3)

Table of Figures

Figure 1.1 Different Overlapping K-Decompositions...11

Figure 1.2 Levels of Organization in Campbell Biology...22

Figure 1.3 Common Definitional Criteria for Levels...26

Figure 1.4 Levels of Organization in Solomon et al.'s Biology...27

Figure 1.5 Formal Hierarchies...36

Figure 1.6 Types of Branching Structures in Hierarchies...37

Figure 2.1 Layer-Cake Account of Levels...45

Figure 2.2 Microreduction Relations in the Layer-cake Account...48

Figure 2.3 Multi-level Mechanisms...62

Figure 2.4 Interlevel Experiments...70

Figure 4.1 Oxidative Phosphorylation...136

Figure 5.1 Rowe's Integrative Levels of Organization...196 Figure 5.2 Rowe's Pluralistic Conception of Disciplinary-Bound Perspectives and Levels. .198

(6)

(7)

Chapter One: The Intuitive Appeal and Ubiquity of

'Levels of Organization'

1.1 Introduction

References to the scientific concept of levels of organization1_{are ubiquitous in both}

philosophy and the biological sciences. The image of the world that the concept evokes posits a vertically stratified structure in which the entities and processes of nature are connected together into a graduated continuity: The things found at one horizontal slice of reality somehow 'make up' or 'are continuous with' the things found at the next slice, and so on. This continuity is often depicted as extending from the basic elementary entities and processes of physics all the way through the biosphere. This image in its complete or abridged form is present in the vast majority textbooks of the biological sciences, introductory as well as advanced, and serves to summarize the basic construal of the natural world whose particular workings scientists seek to uncover by explanation (see also Lobo 2008). Philosophers in turn readily cite this image of the world as a self-evident observation in which to cast some of the biggest questions of our time. Questions concerning the reducibility of natural phenomena to lower-levels of organization (Wimsatt 1976; Burian and Stout 1995; Sarkar 1992; 1998, esp. 53-60; Craver 2007a, Ch. 7; Bechtel 2008, 143-48; Brigandt and Love 2010), whether 'emergent' phenomena exist (McLaughlin 1992; Emmeche et al. 1997; Kim 1999; Stephan 1999a; Korn 2005; Theurer 2014), and the nature of causation outside of physics (Malaterre 2011; Love 2012; Ellis 2012; Hoffmann-Kolss 2014; Franklin-Hall forthcoming) are three well-established areas of philosophical discussion that make heavy use of the concept of levels of organization.

1 _{In what follows, the term <levels of organization> refers to things (levels) posited by a certain claim, even if}

their existence is tentative or hypothetical. When the term <levels> appears without qualifier in the text (e.g.,

of organization or of reality), it will refer to <levels of organization> unless otherwise noted in the text. The

term <'levels of organization'> (with scare quotes) refers to the concept of levels of organization as a theoretical notion discussed by philosophers and scientists.

(8)

Despite this ubiquity, the usage of the concept of levels of organization in science and philosophy is governed mostly by its intuitive appeal, whose justification is often taken as self-evident. As such, the term's precise character and significance is rarely developed in detail. The philosopher William Wimsatt, one of the pioneering scholars that has devoted serious effort to analyzing the concept of levels, observes this as well, saying: “The notion of a compositional level of organization is presupposed but left unanalyzed by virtually all extant analyses of inter-level reduction and emergence” (1994 [2007], 203). Talk of levels of organization in the philosophical literature is quickly replaced or used interchangeably with a range of other distinct ideas. For instance, 'levels' is used as a shorthand reference to, e.g., a systematic dependence between certain properties related by supervenience or realization (respectively, Kim 1998, 1999; Aizawa and Gillet 2009), an epistemic ordering of scientific knowledge or disciplines (Oppenheim and Putnam 1958; Waters 2010), as an ontological thesis about the structure of the world (Oppenheim and Putnam 1958; McLaughlin 1992, 50; Churchland and Sejnowski 1992, 9, 15; Wimsatt 1994[2007], 201-202), or as a combination of several uses (Oppenheim and Putnam 1958; Mayr 1982, 65; Craver 2007a, 170-171). The looseness with which ‘levels’ are used has already called for some to eliminate the term from science (Guttman 1976), and at least minimize its usage in philosophy (Potochnik and McGill 2012; Eronen 2013).

More problematically, philosophers who refer to levels of organization often claim to be simply importing the term as it is used in science (e.g., Kim 2002, 2; Rueger and McGill 2010, 379; Potochnik and McGill 2012, 120). However, the number of sustained analyses dedicated to analyzing how the concept of levels of organization is used in science (and particularly biology) is vanishingly small. Apart from a small number of survey articles detailing a number of issues arising from the usage of 'levels of organization' (Wimsatt 1994[2007]; Kim 2002; Craver 2007a, ch. 5), the concept itself has received almost no direct attention in this regard. This is beginning to change.2_{The philosopher Markus I. Eronen}

2 _{Indeed, the speed at which this is changing continues to gain momentum. Since the submission of this}

dissertation in summer 2014, a number of articles have been published that address 'levels of organization' in a manner parallel to the analysis here. In particular, David M. Kaplan has published a recent (2015) paper that calls specific attention to the lack of scholarly analysis on the concept of levels of organization (rather than other cognates of the term 'levels'). Kaplan, however, also goes further than other mere calls for attention by providing a useful summary of the concept in a number of explanatory accounts in philosophy, including the Hempel and Oppenheim's D-N account, Oppenheim and Putnam's account of microreduction,

(9)

(2013; 2014) has recently published two insightful papers on the levels concept, and Alan Love (2012), along with Ingo Brigandt (Brigandt and Love 2012), also offer an interesting analysis of the scientific use of ‘levels’ in the context of questions about causation and pluralistic explanation, respectively. Other authors, including those mentioned above, mentioned above (Rueger and McGivern 2010; Potochnik and McGill 2012), also point out the lack of such sustained analysis of 'levels' as an impetus for their respective treatments of the concept. Nonetheless the character and significance of the levels concept remains largely an open question.

1.2 Analyzing 'Levels of Organization' in Biological Science

The task taken up in this dissertation will not be to take a position in any one of the particular debates in which levels play a role (e.g., reduction, emergence, and causation). Instead, this dissertation will analyze the concept of levels as it is used in the biological sciences. More specifically, this endeavor will entail explicating3_{the character and significance of the} concept of 'levels of organization' for explanation in the biological sciences. By ‘character’ is meant, very roughly, what scientists take to the term ‘levels’ to mean. The two terms ‘character’ and ‘meaning’ are not interchangeable, however. ‘Meaning’ is a more specialized term in philosophy, and is traditionally used to express the semantic content of a word or

mechanistic explanation, and even contrasting the levels concept with Marr's “levels of analysis” framework. Likewise, Carl Craver has in a new (2015) paper also revisited the levels concept, and builds on his (2007a, Ch.5) analysis of levels by clarifying further his 'defining questions' approach to understanding levels (ibid.; see Chapter 2 for more details), and contextualizing his own mechanistic conception of levels among the many cognates of the levels concept. These two papers come as a particular surprise, given the lack of such papers during the duration of this dissertation's writing, and indeed given the silence on the issue of the levels concept in philosophy generally. Another area of philosophy where the levels concept has continued to 'run rampant' since the submission of this dissertation is the discussion of issues pertaining to non-fundamental causation (particularly top-down causation, higher-level causation, and causation in biology generally). Here, two recent papers (i.e. Hoffmann-Kolss 2014; Franklin-Hall 2014) have thematized levels of organization, albeit indirectly, in their arguments against attempts to articulate higher-level causation (especially in biology). The contributions offered in the analysis of this dissertation is proving to be, if nothing else, a timely choice in topic.

3 _{The term “explicate” here is not meant to evoke any commitment to certain technical uses of the word that}

are sometimes used in philosophy (such as in the sense developed by Rudolf Carnap (1950, 3). Rather, what is meant by the term here is closer to the normal English use of the word, i.e. to make the meaning of the concept clearer. What this precisely entails will be explained presently.

(10)

concept in relation to a thing to which that word or concept refers. Instead, by seeking to understanding the character of the levels concept, the analysis that will be offered here will strive to uncover specific features of semantic content attributed by scientists to the concept as a tool used to aid in explaining phenomena. Concurrently, by 'significance' is meant, roughly,

why the concept of levels is used. The ‘significance’ of the levels concept encompasses the

purposes that are attributed to the concept by scientists who use the term to apply certain ideas in working to explain biological phenomena.

The analysis here will show that the concept of ‘levels of organization’ in the biological exhibits a fragmentary character across different instances of scientific usage. This means that though the concept is capable of clarity precision in given instances, it displays a modest semantic incommensurability across these given instances, which needs to be addressed.4_That is, there are differences in the characterizations of what levels are in given instances such that no singular, common standard is available to compare and contrast all instances of usage of the levels concept. Instead, the semantic content that comprises the character of the levels concept in given instances is determined in a contextualized way, i.e., from the perspective of the researcher using the levels concept (cf. McClamrock 1991). This perspective encompasses a point of view from within a scientific discipline in which that researcher has been trained or is making their particular claim involving the levels concept (see Section 1.2.2 and Chapter 4 for more details on how this is specified).

This contextualized approach to determining the content of a scientific concept stands in contrast to another attitude concerning how to understand the character of a concept, i.e. a

comprehensive approach. A comprehensive approach entails searching for an exhaustive,

singular conception of that concept (here ‘levels’) for all instances of its usage. This often manifests itself in the search for the essence of the concept in question. The ‘essence’ of a concept refers to a feature or set of features that are taken to fundamentally or necessarily designate what that concept is (Robertson and Atkins 2013). The distinction between

4 _{See Chapter 4 for a closer specification of what is meant by “incommensurability” between usages of the}

levels concept. Here 'incommensurability' will be understood in the more mundane sense of “local incommensurability” (Kuhn 1982, 670-71), rather than the stronger sense of the term that imply, e.g., mutual intranslatability of impossibility of ever comparing the content of respective claims.

(11)

contextualized and comprehensive approaches to conceptualizing levels (and particularly its character) will be important for the analysis offered here. In particular, two other philosophical accounts of ‘levels of organization’ in scientific usage, which will be central to developing the main theses of this dissertation, represent polarized positions on how to approach analyzing the character of the levels concept (see Chapter 2). One, the layer-cake account, posits a comprehensive conception of ‘levels’, and represents the classical manner in which philosophers typically explicate the levels concept. Another, the mechanistic account, posits a radically contextualized conception of ‘levels’ that rejects any general import for ‘levels’ can be extrapolated from single instances of usage. The contextualized (but not radically so) ‘fragmentary’ account that will be developed and defended here is someplace in between these two accounts.

At the same time, the analysis here will also show that the levels concept exhibits a minimally unified significance across the instances of its usage. More specifically, it will be argued that another feature of concept usage of science, that concept’s epistemic goal, allows for a more unified (but not comprehensively so) conception of levels to be developed. This aspect of importance attached to a concept's 'epistemic goal' has recently been developed by Ingo Brigandt (2010; 2012) specifically for the purpose of analyzing concepts in the biological sciences that appear to display unavoidable, modestly, i.e. “local” incommensurable variation in their semantic content (see especially Chapter 4). An epistemic goal comprises a set of epistemic values, which motivate the usage of the concept in question. Unlike other elements that comprise a concept’s character, an epistemic goal is not a component of a concept’s semantic content, but rather a feature of how a concept is used by scientists. The epistemic goal motivating the use of the levels concept is to structure explanatory problems that biologists engage in their research. Explanatory problems include issues that belong to constructing explanations for biological phenomenon. These issues may comprise questions concerning, e.g., what is required for an adequate explanation for the respective phenomenon, or more subtle issues regarding the basic characterization of the phenomenon as something for which an explanation is sought. The particular manner in which levels can aid in structuring these problems, and hence contribute to their solutions, is dependent on the specific way that ‘levels’ are used in that instance of usage. These instances of usage will be contextualized in

(12)

claims involving levels, or ‘level claims’, which will be introduced in the next section, and discussed in depth in the later chapters of this dissertation.

1.2.1 Level Claims

Identifying the kinds of claims in which ‘levels’ appear is important, as they serve as a sort of token markers expressing a more overarching understanding of levels. There are two general types of claims in which levels typically appear, namely descriptive and hypothetical claims. Both of these types of claims are closely connected to the significance of the levels concept in biology, i.e. the epistemic goal motivating the use of the levels concept. First, ‘levels’ can be used as a descriptive term made in a descriptive claim about a particular system. In this capacity 'levels of organization' attributes organizational features to a system for the purpose of characterizing what a particular system is. These types of level-claims correspond to what William Wimsatt (1974[2007], ch. 9) calls k-decompositions. According to Wimsatt, using ‘levels’ is closely tied to dealing with the complexity that a biological phenomenon can exhibit, which can interfere with scientific attempts to successfully explain that phenomenon. The use of ‘levels’ is for this reason is often built into the way that scientists basically describe, and basically approach describing, that phenomenon (cf. Burian and Trout 1995). For instance, take the visual system of the fruit fly (see Figure 1.1). A number of distinct biological disciplines may be involved in detailing how the insect visual system works, e.g., (neuro-) physiology, (neuro-) anatomy, electrophysiology, and even evolutionary biology. These level-bound disciplinary perspectives many characterize any given biological system in a number of non-overlapping ways (cf. Winther 2011, particularly Winther's notion of “partitioning”). Furthermore, each of these disciplinary perspectives may even be interested in explaining the same phenomena, such as how sensory information is extracted from the external world by the nervous system, how different cells process specific types of information, and how the visual system mediates flight behavior of insects. However, each of these disciplines will possess very different sets of criteria for how to differentiate the system in question into a set of relevant structures believed to be relevant for explaining the

(13)

respective phenomenon (Lewontin and Levins 2007, 151-2; Love 2008; Winther 2011, 401). Using these various list of criteria, different descriptions will be offered for any given biological system, like the insect visual system, each of which corresponds to one or another disciplinary perspective (cf. Lewontin et al. 1984, 276-7).

Figure 1.1 Different overlapping k-decompositions of a complex biological system. A number of disciplinary perspectives (T)n each view one and the same system (here the visual system of the fruit fly), and construct their

own description of the system (k-decompositions, or K(T)n), resulting in multiple ways of differentiating that

system into parts relevant for the respective perspective. Figure taken from Wimsatt (1974[2007]).

Wimsatt characterizes this state of affairs in biology concerning such “descriptional complexity” as a “conceptual morass”:

(14)

“In biology, at least, the picture is further complicated by another factor – that different theoretical perspectives are not nearly as well individuated as in the physical sciences. Thus, anatomical, physiological, developmental, and biochemical criteria, not to mention paleontological information and inferences of phylogenetic relations and homologies, all interact with criteria of evolutionary significance in the analysis of organisms into functional systems and subsystems. This borrowing of criteria for individuation of parts from different and diverse theoretical perspectives is one of the factors which make functional organization in general and biology in particular such a conceptual morass at times” (Wimsatt 1974, 72).

Roughly, this means that applying the term 'levels of organization' will comprise a package of both epistemic and ontological information nested in a disciplinary perspective. What exactly belongs to this 'package' is determined in a contextual manner. This information results, firstly and most importantly, in a description of the system that differentiates that system into its partitioned units. However, since these claims are made from the perspective of a particular scientific discipline, the description that is offered is also accompanied, often implicitly, by both (1) a set of criteria that specifies why the system is differentiated in that way, and (2) a set of methods and techniques that directly inform the description of the system that is given. Multiple k-decompositions are often available for certain biological systems if they are investigated from different disciplinary perspectives. These different level-claims may or may not result in a similar rendering of the structure and significance of a system for a given case of explanation in science, and ascertaining the similarities and differences between these descriptions is often an issue of substantial discussion in historical biological research (see especially Chapter 5).

Second, ‘levels’ can be used as an operationalized term within a hypothetical claim that postulates a more effective means of searching for a solution to a given explanatory problem. The term 'levels' is operationalized in these cases by an implicit descriptive k-decomposition and by a normative prescription to direct research efforts towards the content of the k-decomposition, i.e. how a system should be studied. The “hypothetical” status of these claims

(15)

captures the tentative nature of these kinds of level-claims as suggestions, rather than concretely explicative claim-statements, about a phenomenon for which an explanation is being sought. Another way to summarize this form of level claims is to consider these claims as heuristic procedures that designate reasoning patterns that scientists use to orient scientific investigation from one discipline to one another, and to nature as well. These types of level-claims correspond to what Lindley Darden (1991, 253) calls “strategies” in scientific reasoning.

“If phenomena to be explained can be put into a hierarchy, a way of producing new ideas (in order to explain the phenomena) is to form hypotheses about the behavior of entities and processes at a different level of organization. If other fields have studied that level, then the interlevel relation may also be an interfield relation. Thus,...using interrelations...includes using interlevel relation[s] when a body of knowledge exists at the appropriate level. If no other appropriate level is known to exist, however, then the strategy “move to another level” is less like the interrelations strategy; the latter postulates a relation between known information. Phenomena may point to the existence of an as yet unexplored level, often at a lower level of organization” (ibid.).

In this capacity, 'levels of organization' fulfill two roles for scientific reasoning used to structure explanatory problems: (1) they represent sources of insight to which scientists are directed in their investigation of a particular problem (cf. ibid., 9), and (2) they offer an opportunity to change the change or modify the problem that is being asked. Understanding what these roles mean in a concrete case will depend on the way that levels are understood in that case, but both will exploit the “package deal” of epistemic and ontological information (described above) that the concept of levels offers. In this way, levels can offer insight into problems by postulating, e.g., new structures and investigative techniques with which to investigate a phenomenon. Similarly, by moving “up” or “down” a level, this can offer a straightforward means of generalizing or specializing the scope of the problem it treats, depending on what is deemed appropriate for that case (cf. ibid., 4.).

(16)

levels concept, will be discussed in depth in the final two chapters of this dissertation.

1.2.2 Wimsatt’s Characterization of Levels

Grasping the character and significance of 'levels of organization' in science is a daunting task due to the wide variation it exhibits across its instances of usage (see especially Section 1.4; Chapter 4). One approach, pursued by Carl Craver, analyzes the levels concept by answering three “defining questions” that together articulate the basic meaning ‘level’ in a given instance (Craver 2007a, 171-172)5_{. The first defining question pertains to the types of things make up} levels in the first place. The second defining question concerns specifying the inter-level relation by which things at putatively different levels are related to one another. The third defining question concerns specifying how a particular item is placed at the same level. Though Craver's defining questions are useful for clarifying the meaning of levels in a

particular instance, it leaves completely open the question concerning what the character and

significance of levels in science generally.6

A more faithful attempt to characterize ‘levels of organization’ as it appears in scientific usage is offered by Wimsatt (1974; 1976; 1994[2007]). The analysis offered in this dissertation is best seen as an attempt to offer a more in-depth analysis of Wimsatt’s characterization of levels, which, though insightful and innovative in its own right, is not very clear. Though Wimsatt analyzes levels as a feature of the world, Wimsatt also emphasizes the significance of the concept for scientific efforts to construct explanation. He offers the following characterization for 'levels':

“[L]evels of organization are a deep, non-arbitrary, and extremely important feature of the ontological architecture of our natural world, and almost certainly of any world that could produce, and be inhabited by, intelligent beings. (This gives levels an almost

5 _{See also Kaplan 2015 and Craver 2015, both of whom refer to the “defining questions” approach in their new}

survey articles to understanding levels.

6 _{Indeed, this approach to grasping the task of analyzing the levels concept seems to favor a strongly}

contextualized perspective for conceptualizing levels, which is Craver endorses in his mechanistic conception of levels (see Section 2.3; cf. Craver 2015).

(17)

Kantian flavor.) Levels and other modes of organization cannot be taken for granted,

but demand characterization and analysis. If I am right, compositional levels of

organization are the simplest general and large-scale structures for the organization of matter. They are constituted by families of entities usually of comparable size and dynamical properties, which characteristically interact primarily with one another, and which, taken together, give an apparent rough closure over a range of phenomena and regularities.” (1994[2007], 203-4, emphasis modified)

This dense passage introduces a number of distinct characteristics of levels of organization. The second sentence of the passage is particularly noteworthy as it highlights ‘levels of organization’ as a proper subject of analysis, detached from an embedding debate. The levels concept, Wimsatt is saying, needs to be analyzed in terms of its inherent usefulness in science as communicating simultaneously several ideas about how the world is structured (i.e. detached from other philosophical debates), and how this structure of the world in turn influences the way that science is organized around it. Wimsatt continues this passage by expressing skepticism for the viability of traditional conceptual analysis in philosophy in capturing scientific usage of the levels concept, saying:

“For anyone who still believes in 'necessary and sufficient conditions' style analyses, I note at least five qualifiers in this sentence – all apparently necessary – that would be difficult at best to deal with, and the referents of these qualifiers are also often disturbingly general, and correspondingly unclear. Note also, that I said that levels 'are constituted by,' not 'are defined in terms of.' Definitional language is notoriously

unhelpful in contexts like these. Broad-stroke characterizations, focused with qualifications and illuminated with examples, are more useful.” (ibid. emphasis added)

This dissertation will take Wimsatt's idea here to heart: that a serious analysis of levels will have to be engaged by investigating the scientific contexts of usage in which the concept seems to play such a prominent role. In this spirit, instead of focusing on classical components of semantic content to gain insight into the concept of levels of organization, e.g. meaning and reference, this dissertation will seek to gain insight into other special features of semantic

(18)

content that are more relevant to the scientific usage of ‘levels’, which belong to the character and significance of the levels concept. These features, which include e.g. the scope of level usage, definitional criteria, mode of presentation, and the epistemic goal of a concept, will be detailed below in Section 1.4 (see also Chapter 4).

1.3 Initial Distinctions

The term 'levels' without further qualifier is, on its own, “multiply ambiguous” (Craver 2007a, 163). What exactly does this mean? There are many terms and ideas that are related to, or even derivative of, the term 'levels of organization'. While some of these offer important insight into how scientists investigate and explain nature using the levels concept, others are only erroneously related to 'levels of organization' and therefore not useful for this analysis. This requires a few caveats regarding terminology in order to avoid confusion in what follows.

As mentioned above, the term 'levels of organization' as used in the biological sciences can encompass both ontological and epistemological connotations. These connotations are sometimes expressed using slightly different terminology, despite falling under the related umbrella term 'levels (of organization)'. Intuitively, a single 'level of organization' is taken to minimally refer to a set of structures and processes within a natural system that share similar features, such as their size, the types and magnitude of forces that govern their interactions, or a compositional relation to a common whole to which these things belong. Multiple organizational levels are hierarchical in the sense that the structures and processes found at one particular level are organized together as subordinate elements to the structures and processes found at higher levels, and in turn are superordinate to the more basic structures and processes organized together at lower levels (for more on hierarchical structures see section 1.5 below; cf. Simon 1962; Wimsatt 1976).

(19)

implicitly considered here: 'levels of analysis'7_{and 'levels of explanation'. Both of these} notions will be taken as epistemic cognates of 'levels of organization' because they rely in a non-trivial way on the hierarchical organization in their domains of inquiry postulated by the ontologically-leaning understanding of 'levels of organization'.8_{This is based on three} observations. Firstly, areas of biological research are often demarcated in terms of the level of organization that they investigate, and correspondingly fall under a certain connotation of 'level'.9_{For instance, 'levels of analysis' are typified by a collection of investigative techniques} and methods that are heavily associated with the structures and processes occurring at a particular altitude of the levels of organization that constitute a particular phenomenon (cf. Craver 2007a, Ch. 5, Sect. 2; Richardson and Stephan 2007). Neuroscience is a multidisciplinary field of biology that exhibits this quite well. Molecular neuroscience, for example, investigates phenomena like mechanisms of neurotransmitter-mediated signaling between synapses, or of biochemical cascades resulting from ion channel behavior. Research in this discipline hence focuses on stereotypical types of structures located a particular (or localized set) organizational level of the nervous system, i.e. synapses, receptors, axons, and other sub-neuronal structures. The investigations of these molecular researchers of the brain use specialized techniques such as gene knock-out studies and pharmacological interventions to analyze, e.g., the role of a specific protein within a biochemical cascade of interest. Such techniques are of little use in investigating phenomena involving structures and processes observed at a different level of organization, such as electrophysiological response properties of a single neuron, or interpreting the significance of an fMRI image study, which observes entire brain regions.

Relatedly, generalizations formulated by these levels of analysis that are used to explain biological phenomena also resemble the hierarchical structure postulated by the levels image of the world, if only via their association with the structures and processes that designate an altitude in the hierarchy of organizational levels, whose terms appear in said generalizations.

7 _{An important exception to what is meant by 'levels of analysis' is discussed in Section 1.2.1.}

8 _{McCauley's (1998) analysis of higher- and lower-level approaches to investigating models of cognition}

exemplifies this point as an implicit understanding of underlying the levels concept. See also McCauley and Bechtel 2001, who implement this further in their “heuristic identity theory”.

9 _{The extent to which this kind of demarcation of disciplines holds is very shaky, and will be taken up several}

(20)

Generalizations formulated using, e.g., single-cell electrophysiological recordings are procured to explain a phenomenon characterized at a corresponding level of organization, like direction-selective response properties of a particular neuron to a specific input stimulus that the organism encounters in an experimental setup. Such generalizations may simultaneously be utilized in the explanation of phenomena occurring at different organizational levels, such as whole-organism behavior, which is observed at the level of the whole organism. Generalizations originating from one altitude among organizational levels that constitute a system, especially in the construction of a multi-level explanation, that are nonetheless related to one another via offering insights into a phenomenon investigated at many levels represent “levels of explanation” (Brooks 2010; cf. Craver 2001; Potochnik 2010).

Levels of 'analysis' or 'explanation' are clearly epistemic in their import, with the former covering methodological or perspectival information about a particular phenomenon and the latter covering explanatory information about a particular phenomenon. Nonetheless, even these distinct meanings are made in relation to a set of natural items, which are postulated by the organizational variety of levels. For this reason, in what follows each of these particular terms will be taken as designating a different mode of application of an overarching concept – that of 'levels of organization'.

A consequence of this caveat is that the concept of levels is used to refer to several distinct (methodological, explanatory, ontological) aspects of a phenomenon, possibly simultaneously. Though probably unintentional, scientific usage of 'levels of organization' seems to exploit this openness of the levels concept (see Section 1.4; Chapter 4). As will be seen, the levels concept can be made clear in any specific circumstance, but there remains the question of whether the concept carries any overarching significance across the different instances of its usage.

1.3.1 Erroneous Concepts of Levels

(21)

of organization', such as when referring to a degree or magnitude of something (level of calcium, level of activity, level of occurrences). Another unrelated 'levels' term is 'levels of processing', which refers to a set of usually processes that underlie a particular activity. This form of usage is frequently found in the neurosciences (see also Craver 2007a, Ch. 5, Sect. 3.2.1), and refers to a sequence of processes that tracks information dispersal between areas of the brain.

One particularly troublesome usage of the term 'level' in philosophical discussions conflates the term 'levels of organization' with David Marr's (1982[2010]) notion of “levels of analysis”, which is a key element to his “tri-level” program for computational neuroscience. This conflation deserves special comment due to the influence of Marr's program in the philosophy of mind, where this erroneous association is particularly rampant (see especially Kim 2002, 1-2; Pylyshyn 1984). Marr's program was meant as a unifying conceptual framework for cognitive neuroscience that advocated treating nervous systems as computational (that is, information-processing) devices. Within this program, three distinct “levels of analysis”10_{demarcate how any given cognitive function can be analyzed. These} included the level of computation, level of algorithm, and level of implementation. The “computational level” refers to what and why a cognitive system does what it does, and is typically described in abstract, mathematical terms pertaining to information processing. This was considered by Marr to be of primary importance to explaining neural phenomena. The “algorithmic level” pertains to the implementation of the computation-theoretical in terms of a representational input-output model. The final “implementional level” concerns detailing how the former two “levels” are realized by the physical “hardware” of the system that houses them.

Two assumptions underlie Marr's framework, which make it clear that his notion of 'level' is completely different than the 'level of organization' variety. Firstly, each iteration of these

10 _{This sense of 'levels of analysis' is very different from the cognate sense of organizational levels discussed}

above. Though the syntax used to describe both of these senses of the term is identical, researchers who use them typically do not confuse their specific meaning. Marr, for instance, chose this designation and articulated its particular meaning without mentioning other possible meanings. Confusions between Marrian analytic “levels” and 'levels of organization' in the biological sciences are committed most frequently by philosophers of mind. Though analyzing the frequency, and impact, of this conflation would be interesting, it lies beyond the scope of this thesis.

(22)

“levels” each refer to one and the same system as a whole, which precludes the hierarchical ordering of natural systems and subsystems that comes along with references to levels of

organization (particular levels of organization, though related into a whole, designate

completely different structures and processes, which are contextualized within that whole). Marr's different analytical “levels”, on the other hand, are simply three different (though complementary) ways of describing a given system at a single level of organization. Given this alone it seems more prudent to call these Marrian entities “dimensions of analysis” rather than “levels of analysis”. Secondly, Marr's framework is strongly motivated by the claim that neural phenomena are best explained by abstracting away from their physical implementation. The physical “hardware” of the realizing system housing the computational programming is instead best characterized in a substrate-neutral fashion. Indeed, Marr envisioned his tri-level framework as replacing empirical-based, 'wet-biological' explanations in neuroscience:

[G]one [are] any explanation[s] in terms of neurons – except as a way of implementing a method. And present is a clear understanding of what has to be computed, how it is done, the physical assumptions on which the method is based, and some kind of analysis of algorithms that are capable of carrying it out (Marr 1982, 18; emphasis modified).

This leads to what McClamrock (1991) refers to as the “de-contextualization” of a system from its natural setting (1991, 188). Abstracting away from the physical details, as Marr's framework does, is contrary to how scientific claims invoke 'levels of organization', which aim to do the exact opposite: to contextualize a phenomenon to a set of different structures and processes that constitute its workings in nature. This is best seen by looking at how scientists themselves depict levels.

1.4 Depictions of Levels in Biological Textbook

s

(23)

are depicted in science is necessary. Probably the most familiar depictions of levels of organization in the biological sciences are found in the many textbooks available for the biological sciences.11_{This widespread use of the levels concept is a straightforward way of} substantiating the ubiquity of the concept, given its presence in the large majority of textbooks of all degrees of specialization, whether for undergraduate introductory courses or for advanced graduate or postgraduate uses.12_{Moreover, looking at textbook depictions of levels} offers a preliminary explanation for the ubiquity of the levels concept: Scientists take levels to be capable of expressing a wide range of important ideas concerning how phenomena are explained in biology.

Some examples of general textbooks for the biological sciences include Reece et al.'s

Campbell Biology, and Sadava et al.'s Life: The Science of Biology (both now in their ninth

editions), which are most often used in introductory undergraduate courses at universities. These textbooks utilize levels of organization as a fundamental motif, usually in its opening pages, to conceptualize both important features of the biological world and how to study it. Furthermore, they often portray the same number and identity of organizational levels. These include (in descending order): the biosphere, ecosystems, communities, populations, organisms, organs (and sometimes organ systems), tissues, cells, organelles, and molecules. For instance, Reece et al.'s series Campbell Biology, one of the principal introductory textbooks for college undergraduates, prominently features the hierarchical view of the world in the major themes of biology (see 1.4.2 below). Their depiction of, and comments about, the levels concept will be used to structure the rest of this section. Their depiction of levels is given in figure 1.2.

11 _{It first it may be asked: Why would textbooks be an important insight into scientific practice? They are, after}

all, 'merely' introductory in content and presentation, and hardly display the specialized, professional knowledge found in research articles. Textbooks are used to convey a basic and foundational understanding of what constitutes a particular branch of science. The information that is conveyed is a mixture of both theoretical and practical knowledge, and is instrumental in forming a researcher's initial contact with the body of knowledge attached to a particular branch of science or a specific discipline of that branch. For this reason, textbooks offer a special kind of insight into what science “is”, or at least is taken to be, by students , aspiring scientists and established researchers. In this capacity, textbooks should be seen as scientific tools whose influence is present from the beginning a person's scientific education through one's specialization and accreditation as a researcher in their own right.

12 _{Some may disagree with this observation (cf. Eronen 2014). Substantiating this observation in detail will not}

be pursued in this dissertation. A more in-depth substantiation of the presence of the levels concept in scientific literature nevertheless represents an important follow-up study to be carried out at a later time.

(24)

Figure 1.2 (a – above; b – below) Levels of organization as depicted in Reece et al. (2010). See text for details.

(25)

Reece et al.'s application of the levels concept, however, is decidedly open to interpretation. They introduce levels with the following passage:

“The study of life extends from the microscopic scale of the molecules and cells that

make up organisms to the global scale of the entire planet. We can divide this enormous

(26)

added).

The highlighted terms and phrases in this passage reflect several features frequently attributed to the levels concept, which help to highlight what elements belong to the character of 'levels' in any given instance, and what kinds of things the levels concept is used for. What quickly becomes apparent is the number of distinct ways that levels are characterized in these depictions; i.e. there are different ways that the character of ‘levels’ can be specified. This openness in the way that levels are characterized in these textbooks shows that the levels concept represents the package deal of sorts mentioned above: That is, ‘levels can be taken as meaning several different things, sometimes simultaneously.

1.4.1 The Character of 'Levels' in Biological Science

The above passage from Reece et al. (2010) points to a number of elements that present themselves as comprising the character of the levels concept as given by various conceptions of the term. Three important elements include the scope of the conception, the definitional criteria used to identify levels, and the mode in which the concept is presented.

Scope of 'Levels'

The first notable element of the character of the levels concept mentioned in the above passage is the scope to which levels are applied in nature. ‘Levels’ typically extend to nature in either a global scope or a local scope. Global conceptions of levels are said to extend to the entirety of nature, and thereby encompass all natural phenomena. Reece et al.’s depiction of levels in Figure 1.2 is hence global, as it “extends from the microscopic scale of the molecules and cells that make up organisms to the global scale of the entire planet” (ibid.). Local conceptions of levels in contrast extend only to a limited part of nature, where this extension is determined in a contextualized manner. One clear example of local levels is found in the structure of proteins, which are composed of four well-defined levels of organization (primary, secondary, tertiary and quaternary). The locality of these levels is made clear by

(27)

Alan Love (2012) who discusses protein structure as an exemplary case of levels of organization:

“The hierarchical representation of four structural layers of organization is applicable to proteins and nucleic acids. This categorization of hierarchical organization is extremely robust and well established empirically. But it quickly loses its significance when applied across the spectrum of biological macromolecules. In this sense, the 'levels of structural organization' are localized to a particular domain of inquiry (proteins and nucleic acids) and not reified into a nominalized designation (the tertiary structural level of biological macromolecules)” (Love 2012, 117).

Global and local conceptions of levels are often taken to be mutually exclusive of one another (see, e.g., Craver 2007a, 191; Potochnik and McGill 2012), though the extent to which this is justified will be taken up in later chapters.

Definitional Criteria

The passage from Reece et al. (2010) simultaneously refers to two distinct definitional

criteria with which to identify levels and distinguish them from one another: scale and composition. Scale, i.e. size scale13_{, is used at the beginning and the end of the passage to} identify constituents that designate three distinct levels (molecules, cells, and the biosphere), while another level (organism) is clearly identified in terms of two compositional parts belonging to distinct levels (again, molecules and cells). These two different criteria are typically used to specify the interlevel relations in a hierarchical layout of levels. These are illustrated above in Figure 1.3. The first, composition (Figure 1.3a), specifies a particular element of a system (for instance, a muscle cell) as being a part of an embedding whole (such as a heart). This way of characterizing interlevel relations is central for the main philosophical accounts of levels, which will be discussed in Chapter 2.

13 _{The quality of size for the scale criterion here is directly implied by the terms “microscopic” and “global”.}

Size need not be the only quality designated by the scale criterion, and indeed could instead be an indicator for another quality, such as temporal scale, magnitudinal scale

(28)

Figure 1.3 Two definitional criteria for levels. (a) A set of organizational levels demarcated by composition. In

these types of representations, level identity, including the properties and entities that belong to a particular level, is defined in terms of a thing's parthood in an overall system. Figure taken from Craver 2007a, 194. (b) A set of levels (here, the levels of the nervous system) demarcated by scale. Figure taken from Churchland and Sejnowski 1992, 9.

The second criterion often used to define levels in scientific cases is scale, particularly size scale (Figure 1.3b). This way of demarcating levels offers an extremely accessible means of representing biological systems, which some of the several intuitive features of the hierarchical ordering that levels are often taken to express. For instance, it captures nicely the idea different organizational levels manifest different types of causal relations, and that these causal relations are exercised by typical kinds of entities.

Reece et al. shift between these two criteria in their descriptions of the major levels in biology depicted above in Figure 1.2. There, for instance, ecosystems “consist of all the living things in a particular area”, but also designate the components of another (higher) level, as “[a]ll of Earth's ecosystems combined make up the biosphere. Likewise, organs are identified as “a body part consisting of two or more tissues” (Figure 1.2b), organelles comprise “the various functional components that make up cells”, and the constituent of the bottommost organizational level in biology, molecules, are defined as a “chemical structure consisting of two or more small chemical units called atoms” (ibid; emphasis modified). In each of these instances the referenced levels (biosphere, ecosystems, organs, organelles, molecules) are identified in terms of their internal composition, or their role in their compositional relation to

(29)

a (whole) object at another (higher) level. However, scale is also used, simultaneously with composition, to characterize different levels of organization. Regarding tissues, Reece et al. say that “our next scale change [downward from organs] to see a leaf's tissues requires a microscope...At this scale, we can also see that each tissue has a cellular structure”14 _(ibid. emphasis added). Molecules are similarly identified interchangeably with the scale criterion: “Our last scale change vaults us into a chloroplast for a view of life at the molecular level” (ibid; emphasis added).

Figure 1.4Levels of organization in Solomon et al. (2010). See Text for Details. Image taken from Solomon et al. 2010, 6.

14 _{This is also the first mention in Reece et al.'s depiction of levels that cites a particular scientific instrument in}

(30)

Solomon et al.'s (2011) Biology, one of the competitors to Reece et al.'s Campbell series in the introductory college textbook market, likewise introduce levels of organization as a basic motif of the biological world, saying that “[w]hether we study a single organism or the world of life as a whole, we can identify a hierarchy of biological organization” (2011, 6; emphasis added). Their depiction of levels is shown in figure 1.4.

Unlike Reece et al. (2010), Solomon et al. only mention a compositional interpretation of organizational levels. Each level is defined as either “being made up of” or “consisting of” things located at its lower adjacent level (Solomon et al. 2011, 6). Despite this appearance of offering a univocal, generalized definition of levels of organization, Solomon et al.'s series of textbooks backfires as an attempt to clarify the meaning of levels of organization. Choosing only one definitional criterion for identifying levels does not entail that the levels concept they articulate is any clearer than in other contexts that that utilize several criteria. For instance, there is no clarification for how the part-whole relation that holds between levels (in virtue of composition) could be specified in order to hold generally across all levels for all biological phenomena.

Mode of Presentation

Another element that makes up the character of the levels concept in biology is the mode in which the concept is presented, i.e. that which the concept is taken to express. Looking again at the emphasized phrases in the above passage, the text strongly implies both an epistemic and an ontological mode in which levels of organization are presented. Though the level-bound entities in the passage (cells, molecules, organisms, biosphere) are identified an undeniably ontological way (i.e. are posited as natural entities in the world), the “study of life”, i.e. the science of biology, is strongly implied to emulate the hierarchical layout of the natural world in some way. This becomes clear by looking at how Reece et al. embed several implicatory claims into the passage that the disciplinary structure of biology follows the hierarchical structure of life. For them, “[t]he study of life extends from” the lowest levels of the living world to the highest ones” (Reece et al. 2010, 3). This seems to imply interdependence between the study of life and the “enormous range” of natural phenomena

(31)

(molecules, cells up to the biosphere) divided into “levels of biological organization”. That is, the disciplinary structure of biology (epistemic) appears to mimic (ontological) biological organization. Ernst Mayr (1982) observes this trend in introductory characterizations of biology, saying:

“The formation of constitutive hierarchies is one of the most basic properties of living organisms. At each level there are different problems, different questions to be asked, and different theories to be formulated. Each of these levels has given rise to a separate

branch of biology: molecules to molecular biology, cells to cytology, tissues to

histology, and so forth, up to biogeography and the study of ecosystems. Traditionally,

the recognition of these hierarchical levels has been one of the ways of subdividing biology into fields. To which particular level an investigator will turn, depends on his

interests.” (Mayr 1982, 65)

This point should not be overemphasized, as positing a strong correspondence between nature and science is extremely problematic (Section 2.2.4; Section 3.2). Indeed, Reece et al. (2010) subsequently resists dividing biology itself along well-defined disciplinary boundaries that follow levels, and chooses rather to describe the study of biological phenomena from a more ecumenical perspective throughout the book. Nonetheless, the dual presence of ontological and epistemic modes attached to levels is continuously seen in the interspersing of names of textbook's chapters and units with terms that continuously shift between these ontological and epistemic connotations introduced by the discussion of the organizational levels of nature.15

This open treatment of global levels is recapitulated in other general textbooks as well. Like Reece et al., Sadava et al.'s (2008) Life: The Science of Biology implies a dual ontological-epistemological mode of applying levels of organization by stating that “[b]iology is studied

at many levels of organization” (2008, 8, emphasis added). Like Solomon et al., on the other

hand, levels are defined by Sadava et al. exclusively by their compositional relations between

15 _{The chapter structure in the textbook's table of contents illustrates this to a small degree. For instance, some}

chapters are labeled according to the natural phenomena that are described (metabolism, the nervous system, the cell, photosynthesis), while others chapters and units are introduced with terms referring to the scientific discipline that investigates some areas of the biological world (genetics, “the chemistry of life”, ecology).

(32)

objects at lower levels, as the presence of phrases like “consists of” and “made of” once again make an overwhelming appearance.16

The conception of levels of organization found in general textbooks exhibits an extremely open character. This is visible in the criteria by which levels are characterized, and in the modes in which levels are applied.17

1.4.2 The Significance of 'Levels' in Biological Science

General textbooks like Reece et al. (2010) rely substantially on the concept of levels of organization to articulate the “major themes” of investigation and explanation in biology.

Campbell Biology is exemplary on this point, and their preamble to introducing levels clearly

sets out such a task for the levels concept:

“A better approach [rather simply than memorizing facts] is to take a more active role

by connecting the many things you learn to a set of themes that pervade all of biology.

Focusing on a few big ideas—ways of thinking about life that will still hold true

decades from now—will help you organize and make sense of all the information you’ll encounter as you study biology” (Reece et al. 2010, 3).

Looking again at their textbook depictions of levels illustrated in Figure 1.2, the shifting of definitional criteria by which levels are identified and demarcated from each other clearly

16 _{Their depiction of levels is illustrated in Sadava et al. 2008, 8.}

17 _{This survey of the biological textbook literature is at best preliminary, and represents another point of}

follow-up analysis to be pursued at a later time. For one thing, the epistemic significance of biological textbooks has not adequately been established in regards to other kinds of scientific texts such as original research articles, reviews, commentary, or other kinds of manuals. Moreover, the diversity of scientific textbook literature has only been touched on here; though the term 'textbook' seems to provoke distrust in philosophers for being 'merely introductory', textbooks are available for all degrees of specialization and experience in the sciences. Indeed, textbooks (both advanced and general) are routinely available in laboratories and scientific work spaces as references for researchers of all degrees of competence. Finally, discipline-specificity of textbooks represents another aspect in need of investigation, since, among reasons, although there are many issues that are treated in various different textbooks (such as protein folding, natural selection, and oxidative phosphorylation), these different textbooks themselves can vary starkly in their treatment of the respective issue. Each of these aspects of textbooks as kinds of scientific textbooks bears directly on questions concerning the character and significance of the levels concept.

(33)

resists saying too much concerning the particular character of levels. Instead of committing to one particular, comprehensive understanding of how 'levels' can exhaustively structure the world, the purpose of 'levels' in this and similar passages is to introduce in a basic fashion how problems of biological phenomena are constructed. The features of biological systems that biologists investigate are too nuanced to be captured by accounts that seek to offer a comprehensive conception of levels. Such features include “emergent” properties, pluralistic approaches to explanation, and the perspective-embedment of biological explanation. Ironically, this conceptual openness will also be the reason why philosophical conceptions of levels will not be sufficient to aid in constructing actual scientific explanations, as will be seen later (see Chapter 4, Chapter 5).

“Emergent” Properties

The first major theme of biology introduced by levels is the widespread presence of “emergent”, or rather, holistic properties in nature.18 This means that certain properties are manifested only by things located at a particular level, but not at other levels. This idea con be expressed using different definitional criteria for characterizing levels: “Emergent” can, on the one hand, comprise properties that are possessed by some things that are treated as whole, which are not by the parts out of which it is composed (McLaughlin 1992; Stephan 1999b). At the same time, it can also refer to properties that are simply 'located' at a particular levels, such as explanatory properties possessed by the generalizations, laws, or models that encompass the Regardless of the specific conception of levels one has, the existence of these holistic properties does not mean that lower levels are not relevant to understanding such behavior in biological systems, as the following passage makes clear:

18 _{Emergence remains a contentious issue in philosophy (McLaughlin 1992; Kim 1999; Stephan 1999a, 1999b;}

Theurer 2014). However, the term “emergent” here in fact expresses a much more innocuous conception of emergence, at least in philosophical sense of the word. In particular, the term here expresses only a “weak” form emergence that claims only the existence, and significance, of systemic properties, and is perfectly compatible with a materialistic universe (Stephan 1999b, 50). “Strong” emergence, on the other hand, asserts in addition the inability, in principle, to reduce the emergent property in question to the behavior of its constituents. That is, it is impossible (in principle) to account for the property in question by looking at the constituents of the system that manifests this property. Strongly emergent remains controversial, because it seems to violate the basic premise of materialism. For this reason, strong emergence goes much further in that it is also a metaphysical thesis, rather than only an epistemic one in the case of weak emergence. The term ‘holistic’ is preferable for exactly this reason.

(34)

“[E]mergent [i.e. holistic] properties are due to the arrangement and interactions of

parts as complexity increases. For example, although photosynthesis occurs in an intact

chloroplast, it will not take place in a disorganized test-tube mixture of chlorophyll and other chloroplast molecules...Emergent [i.e. holistic] properties are not unique to life. A box of bicycle parts won’t take you anywhere, but if they are arranged in a certain way, you can pedal to your chosen destination. And while the graphite in a pencil “lead” and the diamond in a wedding ring are both pure carbon, they have very different appearances and properties due to the different arrangements of their carbon atoms. Both of these examples point out the importance of arrangement. Compared to such

nonliving examples, however, the unrivaled complexity of biological systems makes the emergent [i.e. holistic] properties of life especially challenging to study.” (Reece et al.

2010, 3)19

The issues that holistic properties present to the study of biological phenomena demarcate decidedly different sets of questions than those engaged by philosophers who investigate emergence. In particular, this passage articulates several aspects of how systems are conceptualized as explanatory objects of biology. This passage also contains orienting comments that contextualize such properties in natural systems, understood hierarchically, as

biological. Specifically, the possession of holistic properties by non-natural artifacts (like the

bicycle in the above passage) make clear that (1) biological systems are not ontologically distinct from non-living systems, (2) the organization displayed in these systems possess a higher degree of complexity that distinguishes them decisively from non-living systems.

Pluralistic Approaches to Explanation

'Levels of organization' are used additionally to endorse a pluralist approach to explaining

19 _{Comparable statements to emergence are easily found in other textbooks as well. For instance, Solomon et al.}

(2010) also speaks of these holistic properties, also under the guise of emergence, saying: “[t]he whole is more than the sum of its parts. Each level has emergent properties [highlighted as a key term], characteristics not found at lower levels” (6; see also Reece et al. 2012, 3; Korn 2005). Ecological textbooks are especially rife with references to both levels and emergence.

The concept of levels of organization in the biological sciences